www.HIMSSConference.org
#PrecisionHIT
FEBRUARY 11, 2019
ORLANDO, FL
Moving Genomics to General Practice
2
Moving Genomics to General Practice
Nephi Walton MD MS FACMG
Genomic Medicine Institute
MyCode Return of Results Program
200,000+ consented
140,000 sequenced
1000+ results
returned
Return results for all
ACMG reportable
conditions
4
Precision Health Pilot
Preventative measure
1000 patients sequenced as a routine part of clinical care
Offered through patient’s primary care provider
Return results for all ACMG reportable conditions
5
Genetics by the Numbers
46 Chromosomes
~20,0000-23,000 genes
~3 billion base pairs in the human genome
BRCA2 is 83,736 base pairs
Every time you have child they will incur 100 to 200 new mutations
Mutation rate increases with the age of the father
6
Genetic mutations do not usually give you super powers
7
What your genes can tell physicians
What diseases you are likely to develop
How you will react to medication
How you will respond to infectious diseases
How you will behave
How you will sleep
What you eat
8
What is Medical Genetics/Genomic Medicine
Started in the late 1940s
Did not require an MD PhDs were often practitioners
Dysmorphologists Syndrome identification
Familial risk identifiers
We did not even know how many chromosomes we had until 1956
First disease gene mapped in 1983
9
Pop Quiz! -
In what decade was the following statement printed?
“Of all the applications of medical genetics, the one
potentially most valuable is in the field of
preventative medicine.”
10
Pop Quiz!
In what decade was the following statement printed?
“Of all the applications of medical genetics, the one
potentially most valuable is in the field of
preventative medicine.”
Answer: 1946
11
The Role of a Clinical Geneticist
Diagnosis
Management
Prevention
Risk Assessment
Counseling
12
Challenges in Clinical Genetics
More than 6000 known genetic disorders
Nationwide shortage of clinical geneticists
Getting more referrals from specialists
Shortage of genetic counselors
13
Challenges in Clinical Genetics
Healthy people getting direct to consumer tests
MTHFR challenge
Poor reimbursement for testing and services
Uncertainty in classification of variants and diagnosis
Poor access to data that is crucial in our field
Up to one year wait times to see a geneticist
14
Physicians outside of clinical genetics are not
prepared to understand, deliver or manage
genetic test results
15
Systems that allow physicians access to quick and
accurate knowledge about genetic conditions are the
only way to solve this information overload
No one can acquire and maintain the knowledge to
diagnose and manage over 10,000 diseases.
16
Genomic medicine will move from a specialty
which did not even require an MD degree to
become an integral part of practice that is
required of everyone with an MD degree
17
Found to have a pathogenic variant in BRCA1
Elected to have a bilateral mastectomy
18
Discrepancies in Variant Classification Between 9 Laboratories
Amendola, Laura M. et al. “Performance of ACMG-AMP Variant-Interpretation Guidelines among Nine Laboratories in the Clinical
Sequencing Exploratory Research Consortium.” American Journal of Human Genetics 98.6 (2016): 10671076. PMC. Web. 8 June
2017.
19
Concordance of Variant Classification for BRCA1 and BRCA2
Gradishar William.et al. “Clinical Variant Classification: A Comparison of Public Databases and
a Commercial Testing Laboratory.” Oncologist. 2017 Apr 13. pii: theoncologist.2016-0431. .
20
BRCA1 and BRCA2 are two of the
most well studied genes in the
Human Genome…
AND WERE EVEN SCREWING
THOSE UP!
Also Keep in mind the people
classifying these variants are the
“experts” not general practice
physicians
21
Elisha Cooke-Moore vs Curry County Health
Elisha Cooke-Moore vs
Curry County Health
Bilateral mastectomy and
hysterectomy based on
Incorrect interpretation of
a genetic testing report
by clinician
1.8 Million dollar lawsuit
22
The discordance in variant
classification we have now is
primarily based on testing done
on people with disease
When you start testing healthy
people it’s a completely different
ball game!
FGFR3 Mutation - Achondroplasia
23
Exome Sequencing
Now a common clinical test
Reports on the expressed portion of the human genome
Generates data on all your expressed genes
Resulting data is 6-8 Gigabytes of crucial health information
What does the physician get?
24
25
Genomic Data Life Cycle
Sequencing is performed on one individual
Exome generates 6 to 8 GB of data
Whole genome generates 100 to 200 GB of data
Data is processed to produce a list of candidate genes changes
Further refined to a list of reportable variants put into a PDF
So what happens to the other 6GB to 200GB of data?
26
27
MRI Analogy
VS
28
PACS System for Genomics
Data is too large for EHR data structures
Different kind of data requires a specialized viewer
Frequently changes new data added
Requires specialist interpretation
May be useful for any physician to view especially at point of care
29
Integration of genomic data into the EMR for
patient management and decision support is
crucial to enabling general practice
physicians to practice genomic medicine
30
Summary of Challenges
No way to store genetic data
Clinical Genetics workforce will not scale to handle sequencing everyone
No way to utilize genetic data in a clinical workflow
Clinicians outside genomics not equipped to manage genetic testing results
31
Genetic Data What to Store
1. Discrete Variants with Known Health Related Effects
1. Point Mutations, Indels
2. Copy Number Variants
2. Filtered Data
1. VCF
3. Raw Data
1. BAM, FastQ
32
Genetic Data Where to Store
1. In EHR
2. In External Database linked to EHR
33
Discrete Variants
Currently reported as in a PDF
Workarounds include creating lab values that store variants
Only in past year has ability to store discrete variants been available
34
Getting Genetic Variants into the EHR
Epic has recently implemented “Genomic Indicators”
Will allow for clinical decision support around genomic variants
Gives genomics a home in the EHR finally a place to put the data
35
Consortium for Agile Genomics Getting Data From The Lab
36
HOLD UP!!!!
PDF from the lab ok to put into the system automatically
Discrete variant information as a genomic indicator NOT OK
Why?
37
Bad Things Can and DO Happen
Case
Patient in her 20s evaluated for connective tissue disease
Clinically evaluated as hypermobile EDS
Aortopathy panel sent
VUS in gene for Loeys Dietz
Cardiac MRI normal
38
Case continued
Patient seen by Ob/Gyn provider
Provider sees genetic test
Assigns diagnosis of Loeys Dietz in Epic
Patient instructed that she MUST have a C-section
Schedules monthly echocardiogram
Patient calls genetics department
39
Providing too much information or too little
Providing information that could be used the wrong way potentially exposes
the patient to unnecessary and potentially harmful action.
Hiding potentially useful information prevents patient from receiving the
appropriate action.
Where is the balance?
40
If malpractice is any indication
Only a handful of cases regarding taking action because of wrong
interpretation.
Versus.
57% of all genomic medical malpractice cases could have been avoided if
genetic testing had been performed
41
Pharmacogenomics
New clinical exome laboratory offers PGx variants as option to patient.
General response - Yay!
My response Uh Oh!
Multiple patients with pharmacogenomic results
Who counsels the patient
Is there a liability associated with not acting on results
How do physicians find the report?
42
The rest of the DATA!
Which data is used?
Raw Data
Filtered Data
Annotated Data
Copy number call from exome/genome
43
Storing – “The rest of the data”
No place to store VCF files
No place to store discrete phenotypic information
A Genotype- Phenotype Archiving and Communication System (GPACS)
NIH SBIR grant in partnership with PhenoSolve (SimulConsult)
Grant Number NIH 1R43HG010322-01, PI: Segal, MM
44
GPACS
45
Phenotype Builder
Extracts phenotypic elements from free text and ICD10 codes
Builds phenotypic profile of HPO terms
Stores HPO terms in GPACS
46
We Have Genetic Data in the EHR Now What?
Make sure genetic data is used accurately/effectively
Diagnosis
Clinical Decision Support
Information Delivery to Providers
Information Delivery to Patients
Take Actions
47
Use of Genomic Information in Clinic - Simulconsult
48
eMERGE PGx
Clinical Decision Support built for
CYP2C19 and Clopidogrel
SLCO1B1 and Simvastatin
49
Delivering Information to Patients and Providers
Working with Epic to implement this with genomic indicators
Infobutton will reside next to the genomic indicator
Provides information related to gene or variant
50
Compass/Gene Report
HL7 compliant web service
Activated through InfoButtons or
direct web-link
Delivers different information based
on user - patient vs provider
Delivers information specific to
variant, sex, and age
51
Implementation of Chatbots
Partnership with Clear Genetics
Use Chatbots for:
Consenting patients
Following up on test results
Patient questions
Cascade testing
Chatbot work by Amy Sturm and Tara Schmidlen
52
Consent Chatbot
53
Patient Follow-Up Chatbot
54
Family Sharing Tool
55
Cascade Chatbot
56
Smart FAQ
57
What’s Next?
How do we integrate all genomic information into the workflow?
Which variants do we include and who decides?
When working outside an ”expert” whose consensus guidelines are used?
How can we effectively use information in real-time to impact patient care?
58
Acknowledgements
Geisinger
Marc Williams
Darren Johnson
Bruce Levy
Dean Parry
Jordan Olson
Bonnie Spandra
Rebecca Pulk
Chancey Christenson
Amy Sturm
Tara Schmidlen
University of Nebraska
Scott McGrath
Simulconsult
Michael Segal
Lynn Feldman
HPO
Peter Robinson
Epic
Jordan Dantas
InfoButtons
Guilherme Del Fiol
HL7/FHIR
Gil Alterovitz
Grant Wood
59
QUESTIONS?